Purification and structural characterization of the putative gag-pol protease of human immunodeficiency virus.

CopyrightX 1988, AmericanSociety for

Microbiology

Purification and Structural

Characterization

of the Putative

gag-pol

Protease of Human Immunodeficiency Virus

ERIK P.

LILLEHOJ,l

F. H. RICK SALAZAR,2 ROBERT J. MERVIS,3 MICHAEL G. RAUM,4 HARDY W. CHAN,2

NAFEES AHMAD,3 ANDSUNDARARAJAN VENKATESAN3*

Laboratory ofMolecular Microbiology3 and BiologicalResourcesBranch,4NationalInstitute ofAllergyandInfectious Diseases, Bethesda, Maryland 20892;ProgramResources, Inc., Frederick Cancer ResearchFacility, Frederick,

Maryland 217011; andInstituteof Bio-Organic Chemistry, SyntexResearch, Palo Alto, California 943032 Received 28 January 1988/Accepted 25 April 1988

We have purifieda 10,774-daltonprotein from human immunodeficiency virus (HIV)type 1 that isencoded in the protease domain of thepolopen reading frame (ORF). Radiochemical amino acid microsequencing identified 12 aminoacids from the stretch of 39 N-terminal residues ofthisprotein,beginning withaPQITLW

sequenceatposition 69 ofthepol ORF. Radiosequencing of selected tryptic peptides oftheproteinidentified 11additional residues (Leu-9 and Val-2) in six peptides encompassingtheentiremoleculeof 99 residues. A protein of similar sizeand identical N-terminalsequence(determined through the first 39 residues)waspresent

amongtheprocessed HIVpolgeneproducts in Escherichia coliwhich expressed the entire HIVpolORF. The

Cterminus ofboththe viraland E. coli-expressed proteinswasinferredtobecontiguous with the N terminus of the p64-p5l reverse transcriptase onthe basis oftryptic mapping and specific immunoreactivity withan

antiserumagainstadodecapeptide locatedupstreamof thereversetranscriptase. Thus, theinitial processing ofthepolprecursorthatgeneratesthe native proteaseisapparently preservedacrossphylogenetic barriers.

Although the purified viralproteaselacked measurable proteolytic activity, the bacterialextractswerecapable

ofprocessinganHIVgagprecursorprotein synthesized in E. coli.

Genetic and biochemical studies have established the requirement ofa specific viral protease for the expression andprocessing of the gagand polgeneproducts of several

retroviruses (4, 7, 19, 20).Among the retroviruses studiedso

far, a viral protease is synthesized as part of the gag or

gag-pol polyprotein, encoded either within the gag orpol

openreading frame (ORF)orbyaseparateORFoverlapping bothgagandpol (10-12, 15, 21). Viralproteasesof 13 and14 kilodaltons (kDa) have been purified from murine leukemia virus and bovine leukemia virus and extensively character-ized(21, 22).

Manyof the mature internalstructuralproteins of human immunodeficiency virus (HIV) type 1 are derivedfrom the

proteolytic processingoftwoprimary translation products of 55 and of 180 to 200 kDa corresponding to the gag and

gag-pol polyproteins, respectively (8, 9, 13, 16).Two forms (p64 andp51) of HIVreverse transcriptase (RT)and a p32

protein(the putativeviralintegration protein) juxtaposed to the C-terminal side of the p64 RT have already been

se-quenced (13, 17, 18). Although acandidateproteinencoded

bytheN-terminal 167 residues of thepolgenethatoverlaps the gag ORF by 82 residues had not been discovered in isolated virions orinfected cells,thestructural homologyof this domain of thepol ORFwith other retroviral proteases prompted us to analyze extracellular virions for viral

pro-teins using antisera against peptides corresponding to the

gagand

pol

ORFs.Byuseoflimitedaminoacidsequencing,

we have identified a ca. 10-kDa protein from the virus

particles andEscherichiacoliextracts expressingthe entire HIVpolORF andpossessinggagprocessing activity. Dur-ingthecourse of thiswork, Debouck etal. havereported a

gag-specific proteolytic activity associated with a 10-kDa protein from E. coli extracts which express the protease domain of the HIVgenome (5).

*Correspondingauthor.

The gag and gag-pol gene products were analyzed by

immunoprecipitation of intracellular viral proteins labeled under steady-state or pulse-chase conditions with either

pooled sera from acquired immunodeficiency syndrome (AIDS) patients orrabbit hyperimmune sera raised against E.colifusion proteins containing discrete structural domains of thegagandpolORFs.Labeledextracellular viralproteins were purified by immunoprecipitation, and their partial

N-terminal sequences were determined. Figure 1A illus-trates the mappositions of these various gag (S.

Venkate-san, unpublished data) and pol (13, 18) gene products. During theseearly studies, asmall(ca. 9to10kDa) protein

was consistently visualized among the immunoprecipitates with pooled sera from AIDS patients (Fig. 1B). It was

occasionally immunoadsorbed by gag antisera and was

presumed to represent a gag gene product. Direct N-ter-minal radiosequence analysisof this moiety revealed equi-molar abundance oftwodifferentsequences. Onewas iden-tified as beginning at position 378 of the gag ORF and

probably representedaprocessed productof the C-terminal

p15 gagprotein (Fig. 1A). The other sequence determined for the 9-to 10-kDaproteinwastentatively localized begin-ningat69 residues from thebeginningof the

pol

ORF.

The 9- to 10-kDa moiety was electroeluted from

acryl-amidegels andchromatographed on aDEAE-cellulose col-umn. Successive N-terminalprotein sequencingof two

pro-tein fractionselutingat0.2 and 0.5 M NaClwasundertaken. The N-terminal sequencethrough 30 degradative cycles of the 0.5 M NaClfractionexactlycoincided with astretch of residues starting at position 378 of the gag ORF (R. J. Mervisetal.,submittedforpublication). Anotherproteinof 9to10kDawaseluted with 0.2 M NaCl. Toavoidconfusion, the 0.2 M NaCl-eluted protein will be referred to as plO, while the gag protein eluting with 0.5 M NaCl will be referred to as p9. Multiple sequence analysis of the plO protein yielded an unambiguous sequence. Figure 2a illus-3053

on November 10, 2019 by guest

http://jvi.asm.org/

A

B

L

kDa M

M,L

M,V

34

6

34_

9f 7

]env

p64

p55

p41

---4

9

DRY GEL AUTORAD/OGRAM

FtUORPOGRAM FIG. 1. Polypeptides encoded by the gag and pol genes ofHIV. ArecombinantHIV strain

generated

previously by

transfection ofan infectiousproviral DNA,pNL432, into SW480 cells(1)wasused in all thestudies.Forvirus

labeling experiments,

100mlofA3.01cellsin RPMI1640medium(106cells perml)wasinfected with1mlof virus inoculum(105to106 50% tissure culture infective doseunits)and labeled with

[35S]Met

and theindicated

3H-labeled

amino acidasdescribedpreviously (13).Theextracellular viruswas

purified by

precipitation

with polyethylene glycol6000 (13). In someexperiments, the virus wasfurther purified by

centrifugal

banding

at

35,000

rpmfor 90 min ina

Beckman SW40rotor at the60% interphase ofadiscontinuoussucrosegradient (60, 35,and20%[wt/vol]in20 mMTris

hydrochloride [pH

7.5], 1mM EDTA[TE]buffer).Theviruswassolubilizedbythe addition ofNP-40and TritonX-100to0.5% and

dialyzed

twice

against

500 volumesof TE buffer with detergentsbeforeelectrophoreticorchromatographic analysis. (A) Schematic illustration of the HIV genome and the relevant ORFs. The viral polypeptides derived from the gag and pol ORFs andvisualized in the infected cellsorpurified virus are

indicated in the lower part. The180- to 200-kDagag-pol precursorwasidentified andmappedwith definedantisera inpulse-chaseexperiments ofinfected cells ortransfectants expressingaseries ofoverlapping gag-pol deletion plasmids(9; S. Venkatesan, unpublished data). My, myristyl groupfound at the Ntermini ofp55, p4la,andp17.Mapping ofthep4laandp4lbgagproteinsis basedonimmunoprecipitationand limited protein sequenceanalysis (Mervisetal.,submitted). Alsonotethat theassigned cleavage siteseparating p9andp7is tentative. The position oftheproteasedomain is alsoshown.(B) Viralextracts(0.4ml)from2 x 108A3.01cellswereincubated for4hat4°Cwith10,ul of pooled antiserum frompatients with AIDS. The immune complexes wererecovered bybindingto protein A-Sepharose, exhaustively rinsed, and eluted for electrophoresis under reducing anddenaturing conditions on15 or 10 to20%gradient polyacrylamide gelsin SDS. Resultsobtainedwith10 to20%gradient acrylamide gelsinSDSareshown.LanesM; M,L;M,V; andLillustrate results obtainedwith virus labeled with therespective aminoacid(s) (single-lettercode).

trates the results

obtained

when plO was labeled

with

Met, Leu, Lys, or Val. The

experimentally

determined occur-rences of Leu at positions 5, 10, 19, 23, 24, 33, and 38; Lys at

positions

14 and 20; Met at position 36; and Val at

positions

11and 32

precisely corresponded

to

their

positions

within a domain of the deduced sequence of the pol gene startingat residue 69. The probability value for the random occurrenceof suchasequence wascalculated (13) tobe less than 1.042 x

10'-,

thus ruling out the possibility that this sequence wasderived from contaminating cellular proteins.

Since direct

N-terminal sequence analysis was reliable for only 35 to 40 degradative cycles, we sought to obtain the internal sequence of the plO protein from its tryptic peptides. Individual peptides labeled with either Met and Leu or Met and Val were purified by reverse-phase high-performance

liquid

chromatography (RP HPLC), and their N-terminal sequences were determined. The results obtained with the Met-and Leu-labeled tryptic peptides are shown in Fig. 2b. Not all the tryptic peptides predicted by the deduced

se-quence of this region of the pol ORF were recovered by HPLC. This

might

have been dueto the

hydrophobic

char-acter of some of the internal

tryptic peptides.

Six Leu-labeled

peptides

that were

ultimately recovered and

se-quenced

are identified within the deduced sequenceof the protease domain. For

instance,

the

peptide

contained in HPLC

peak

L2waslocalized between residues 83 and 88 of thepol

ORF,

that in

peak

L3 was between residues 77 and 82, that in

peak

L7 wasbetweenresidues 89 and 109, that in

peak

L8wasbetweenresidues 69 and 76, and that in

peak

L9 was between residues 114 and 123. The sequence of Val-labeled

tryptic peptides

further confirmed the peptide

assign-mentsshown in

Fig.

2.

The HPLC-eluted viral

protein

had no demonstrable pro-tease

activity despite

experimental

variations

of

protein concentration

(0.8

to 3.6

mg/ml),

pH (2.5 to 7.0), ionic

strength (O

to0.5 M

NaCl),

incubation time(1to48

h),

and temperature

(22

or

37°C),

conditions under which proteases of mammalian

(22)

and avian

(2, 6, 19)

retroviruses are

p16

on November 10, 2019 by guest

http://jvi.asm.org/

[image:2.612.51.555.66.362.2]

a

00% 0

2

0_

0

z

on

('4 0

_-b

8

6

4

2

4

2

2

1.5 1.0

0.5 o 0

U'

1.5 (A

1.0 0

0.5 X

0-%

0

_-1

CYCLE NUMBER

P4

p34

RioPROTEASE

UL6 II I

L2I2

LI7III II I LIO

69 76 2 Ua 109-113 1231-15 136 155 167

69 so 100

. . . .. . .

I'QI'I.WQRPr.VTIKIGGQLKEALLI)T(;ADI)TVIEEIMSI.P(;RWKI'u(MI

1,8 L L2 L7

[image:3.612.166.469.69.468.2]

120 140 167

... ...

I.9 i. ''

FIG. 2. N-terminalsequenceanalysis of the intact HIV protease and its tryptic peptides. Extracellular viral proteins eluted from the gel after immunoprecipitationwerefurther purified by adsorption to a DEAE-cellulose column (1.0 by 5.0 cm) in TE buffer containing 0.5% NP-40 and0.5% Triton X-100, and the column was eluted batchwise with TE buffer plus detergents containing 0.1, 0.2, 0.5, and 1.0 MNaCl.The individual fractionswereanalyzed by SDS-PAGE. Limited N-terminal sequencing of all fractions was undertaken. The 0.2 M NaCl fraction thatyieldedahomogeneous sequenceofthecandidateprotease wasmixed with 1.0 mg of crystalline bovine immunoglobulin G, reduced, alkylated, and exhaustively digested with tolylsulfonyl phenylalanyl chloromethyl ketone-treated trypsin. The tryptic digests were fractionated by RPHPLC (14) on aC18column (Supelco Inc.) with a linear gradient of 0 to 100% aqueous acetonitrile containing0.1% trifluoroacetic acid. Samples were measuredfor radioactivity, and selected peaks wereconcentrated by lyophilization and processedfor N-terminalsequencedetermination. (a)Radiochemical sequencedeterminationofthe plOprotein(0.2 MNaCl-DEAEfraction) labeled with

[35S]Met

and

[3H]Leu

(A),

[35S]Met

and

[3H]Val

(B),or[3H]Lys(C). Viral bandswerevisualized bywetgel autoradiography and recovered by electroelution (14). Automated N-terminal microsequence analysis by sequential Edman degradationwasperformed withaBeckman 890M sequenatorand the Beckman0.1 MQuadrolprogram042386 (3). Onthebasis of thereleaseof the individual amino acids through39cycles, auniqueHIVproteinsequence in theproteasedomainwasidentified(topline). Theoccurrence of

[35S]Met

atcycles1,2, and 10inprofile B wasattributedto aminorcontaminant which constitutedless than5%oftheprotease-specific sequenceonthebasis ofarepetitive yield of96%. (b) Tryptic peptides of the putativeplO HIV protease labeled with

[35S]Met

and either

[3H]Leu

or

[3H]Val

werefractionatedbyRP HPLC, and their partial N-terminal amino acidsequencesweredeterminedandaligned withinthepredictedtryptic peptides(L2, L3, L7,L8, andL10)of theproteasedomain. The residues identifiedby sequencing oftheselectedpeptidesareindicatedbyasterisks.

active.

Asan

alternative,

we

analyzed

theprotease

activity

in extracts

of

E.

coli expressing

the HIV pol

ORF,

since these extractshad

substantial

RT

activity

and

readily

detect-able amounts

of

mature

p64-pSi

and p34 HIV

pol

proteins,

suggesting

that the

pol

ORF gene

product

was

processed

in these

cells,

presumably by

the HIVprotease. The crude E. coli extracts were

screened for

agag protease

activity

with a

purified preparation

ofatruncated HIV

"gag"

precursor

protein expressed

inE. coli.This

synthetic

moleculewas a

lacZ fusion

protein

consisting

of six residues of lacZ fol-lowed

by

a stretch of 348 HIV

gag-encoded

amino acids

starting

with residue 57 of the gag ORF.

Cleavage

of this

"gag"

substrate at the N terminus of the mature

p24

gag

protein

wouldgenerate a ca. 32-kDa

product

containing

the p24 and the N-terminal residues of the distal

p9

gag

protein.

The

enzymatic

reactions wererun for 2or4

h,

electropho-resedon 15%

polyacrylamide gels

insodium

dodecyl

sulfate

(SDS),

and

analyzed

by staining

with Coomassie brilliant

I p- I

6

on November 10, 2019 by guest

http://jvi.asm.org/

1 2 3 4 kDa

B

974

68.0

,43.0

4--gag"SUBSTRAIE

+-gag"

-25.7

- 18.4

[image:4.612.55.296.76.261.2]

-14.3

FIG. 3. Enzymaticassayof theE.coli-expressedprotease.

Bac-terially expressed protease was partially purified from E. coli transformedbyarecombinantpUC8 plasmid carryinga

3,746-base-pair BglII-SalI HIV proviral DNA fragnment (obtained from a

subgenomicHIVproviralDNAplasmid, pBenn2 [1]) containingthe

entirepolandsorORFs. The E. coli transformantscontainingthe

HIVpol ORF fused to the lacZ gene were induced with 1 mM

isopropyl-p-D-thiogalactopyranoside

for4 h and then extracted ina buffercontaining20 mM Trishydrochloride (pH 7.8),1.0 MNaCl, 0.5%NP-40,and 0.5% Triton X-100.After clarification(10,000x g, 10min),thesupernatantswereassayedforRTactivityanddialyzed exhaustively against low-salt (0.1 M NaCl) extraction buffer by using tubingwithacutoff size of 2to3 kDa. ForassayingtheHIV protease,anE.coli fusionprotein containingapartof the HIIVgag ORF was used. A 1,087-base-pair MboI-BgII fragment (corre-spondingtonucleotides 1016 to2098 of the HIVgenome) of HIV

proviralDNAcontainingthe middle one-third of thegagORFwas fusedtothe lacZgeneofpUC8,andprotein expressionwasinduced with

isopropyl-p-D-thiogalactopyranoside.

A 42-kDaE. coli HIV gagfusionprotein containingthe C-terminal 57 residues ofp17,the entirep24, and the N-terminal 60 residues ofp9wasidentified by immunoblottingwithgag-specificantisera. Thisproteinwaspurified fromthe cellextracts tonearhomogeneity bytwosuccessivecycles of RP HPLC as detailed above. The 42-kDa gag protein was incubatedatroomtemperaturewith variousamountsofanextract

ofE. coli whichexpressedthepolORF. All reactionswereanalyzed

by SDS-PAGE and staining with Coomassie brilliant blue (A) or

immunoblottingwithagag-specificmonoclonalantibody (0058)and

125"-labeled

proteinA(B). (A)Lanes: 1 and2,mockassayreactions containing only 15 ,ug of"gag" substrate incubated for 2or4 h,

respectively; 3 and 4,incubationscontaining only 10 p.l ofextract

fromE. coliexpressingthepolORF for 2or4h, respectively;5and 6, completereactioncontaining 10 of E. coli-expressedenzyme

plus15 jigof"gag"substrate incubated for 2or4h, respectively.

Aprominent10-kDa bandseennearthe bottom of thegelin E.coli

extractsexpressingthepol ORFprobably represents theputative

protease. (B) "gag" substrate(5 ,ug)wasincubated for 4 h with 10

(lane 1),2.5(lane 2),or0(lane3),u1 of E.coliextractexpressingthe polORF. The substratealone is shown inlane 4. The positions of

the42-kDa"gag"substrateand the 32-kDa"gag"reactionproduct areindicatedbythearrows. Thereactions in panels A and Bwere

analyzedondifferentgels.

blue(Fig. 3A)andimmunoblotting witha mousemonoclonal

antibody (0058) raised against theHIV p24gagprotein(Fig.

3B). The crude extractswere capable of cleaving the HIV

"gag" substrate, yieldinga32-kDaproduct (Fig. 3A, lanes5

and6,andB, lanes 1and2).Incontrast,neither the purified substrate northecrudeextractalone produced this reaction product. Also, E. coliextractsexpressing thevectorplasmid alone hadnoenzymatic activity.

Partially purified

bacterialextracts

expressing

the HIV

pol

ORF

hada

readily

detectableband of 10 kDa

(Fig.

3A,

lanes 3

through

6). The extracts were fractionatedby RPHPLC, andthe HIV

proteins

in individual fractions were detected

by immunoblotting

with

pooled

sera from AIDS

patients

(Fig.

4A1 and

2).

The bulk of the viral

proteins eluting

with the acetonitrile

gradient

included bands of 64 and 51kDa,as

expected

for theHIV RT.

Indeed,

the fractions enriched for the

p64

and

p51

bands had the maximal RT

activity (Fig.

4A1,

fraction

30).

Two additionalsmaller viral

proteins

of 10 and of 15 to 20 kDa were

preferentially

eluted with an

isopropanol gradient (Fig. 4A2).

These two

proteins

also

specifically

immunoreacted with a rabbit antiserum raised

against

a

synthetic peptide corresponding

tothe C terminus of the

putative

viral

protease

(residues

154 to 167 of the HIV

pol

ORF; Fig. 4B).

None of the individual HPLC

fractions,

however,

had

reproducible

gag

protease

activity

when the assay described abovewas used.

To

obtain

purified

E.

coli-expressed plO protein

for amino

acid

sequencing,

the bacterialextractslabeledwith

[35S]Met

and

[3H]Leu

or

[3H]Val

werefractionated

by

RP HPLC and individual

fractions

were

immunoprecipitated

with

pooled

antisera

from

AIDS

patients

and resolved

by

SDS-polyacryl-amide

gel electrophoresis (PAGE) (Fig.

4C).Two

proteins

of 10 kDa

(Fig.

4C, arrow)

and of 15to20

kDa, eluting

atca.

40% and 55%

isopropanol, respectively,

immunoreacted with the

pooled

sera.Rabbit antisera raised

against peptides

corresponding

toHIV

pol-encoded

residues 71 to 128

(data

not

shown)

or 154to 167

(Fig.

4B)

also

specifically

reacted with these

proteins.

The

plO

protein,

labeledwith Met and Leuor Met and

Val,

was

electroeluted,

and its N-terminal sequence was determined. This sequence was identical to that of the N terminus of the HIV

plO

protein.

Limited sequence

analysis

of the

larger species (15

to 20

kDa)

also revealed the same N terminus

(data

not

shown). Although

the

C terminus

of neither

protein

was

determined,

the fact that

both

proteins

immunoreacted with rabbit antisera

against

peptides

corresponding

to the

pol-encoded

residues between 71 and 128or154 and 167 localized theirsequences withina

99-residue domain

between codons 69 and 167 of the HIV

pol ORF. Neither of

the

proteins

immunoreacted

with monoclonal antibodies

against

the HIV RT sequence

(data

not

shown).

HPLC

fractions enriched for

bacterially expressed

plO

lacked

reproducible

protease

activity, probably

because

of low recovery, poor

stability,

or

extensive denaturation.

The

following

lines of evidence

strongly

support the

candidacy

of

plO

asthe

authentic

HIVprotease.

(i)

Theprotease

activity

could be

detected

only

in E.

coli

expressing

the

HIV

pol

ORF

andnotin E.

coli

carrying

thevector

plasmid

alone.

(ii)

The gag

processing activity

was

abolished

by

antibodies

directed

against

allorpart

of

the protease

domain

(data

not

shown).

(iii)

Bacterial

expression

of

a small DNA

fragment

encompassing

the protease

domain

yielded

a10-kDa

protein

that

processed

gag precursor

(5).

(iv)

pol

ORF

deletions

spanning

the N terminus

of

theprotease or

eliminating

the entire protease domain abolished the appearance

of

the

p64-p5i

HIV

RTconcomitant with the

disappearance

of

the 10-kDa

protein

inE.

coli (R.

Swanstrom, personal

commu-nication).

The

relatively

low

yield

of the plO

protein from

virus

precluded

C terminus

analysis by carboxypeptidase

degra-dation.

Unlabeled, HPLC-purified,

E.

coli-expressed

plO

was not amenable to

digestion, either.

The deduced se-quence of the

pol ORF

would

predict

a 12-residue

tryptic

fragment

between residues 156 and 167 if the

plO

and RT

1 23 4 5 6

A

Ak.

qw#.

F"'I

0,

on November 10, 2019 by guest

http://jvi.asm.org/

50 55 60 65 70

B 1 2 3 4

5

kDa

- 46.0 - 30.0 - 21.5

do

- 14.3

40 0

o

.0

20

ir

so₀

50l

o0

a

s

C

IHPLC 45 46 47 48 4950 51 52 53 54 55 56 57 58 59 60 6152 FRACTION

.Ifls

_ _ ____

.,s_

_

/-0

i.

1 69 168 1015

pol ORF H2N-FFRED-PQITLWORPLVTIKIGGOLKEALLDTGADD-- PISP-- ED-COOH A AA A AA

FIG. 4. Purificationand amino acidsequencedeterminationofE.coli-expressed protease.Thedialysateof HIVpolORF-expressingE. coliextracts(asdescribedin the legendtoFig. 3)wasappliedtoa20-mlcolumn of DEAE-cellulose in0.1 MNaCl, and theproteinswere

elutedby batchwisesalt gradients of 0.2, 0.5, and1.0 MNaCl. AfterRTanalysis and immunoblotting, theplO-enriched0.1to0.2 MNaCl

fractionswereconcentrated byacetoneprecipitation, dissolved in6 Mguanidine hydrochloride containing 0.1%trifluoroacetic acid(TFA),

andresolvedby RPHPLCon aVydac C4 column developed withalineargradient of 0to30%aqueousacetonitrile-0.1%TFAfor 30min followedby isocraticelution for10minanda60-mingradient of30to60% acetonitrile-0.1%TFA.Thiswasfollowed by isocratic elution for

10min with20%aqueousisopropanol-0. 1%TFAanda20to100%linear gradientofisopropanol-0.1% TFA for60min.Fractions(2 mleach)

werecollected, and samples of selected fractionswereelectrophoresed and immunoblotted with pooledserafrom AIDSpatients. (Panels Al

andA2)SDS-PAGEprofiles of E. coliextractswhichexpressthe UIIVpolORF resolved by RP HPLC. The discontinuous linear acetonitrile

(Al)andisopropanol (A2) gradientsaredenoted by lines drawnacrossthe panels. The proteinswereelectroblotted and screened with pooled

serafromAIDS patientsand 1251I-labeled protein A. Selected fractionswereassayedforRTactivity,and the RTprofile(fractions 25to34)

isillustrated byagraph in panel Al. (B) Immunoblot detection of the twoforms oftheputative HIVprotease expressed inE. coli.Two separateisopropanol gradient fractionscontaining thetwoforms of the putative HIVprotease(isopropanol gradient fraction58, lanes 2and

4; isopropanol gradient fraction64, lanes 3 and 5) were lyophilized and rerun on 17.5% polyacrylamide gels in SDS, electroblotted to

nitrocellulose,andimmunoreacted witharabbit hyperimmuneserumagainsta syntheticpeptide containing residues 154to167of the HIV

pol ORF (lanes 2 and 3) or pooled sera from AIDS patients (lanes 4 and 5). The isopropanol fractions 58 and 64 were mixed and

electrophoresed (lane1)and reacted with nonimmune rabbitserum.(C)Immunoprecipitation ofradiolabeled bacterialextractswithpooled sera from AIDS patients. A 100-ml sample ofanE. coli culture expressing the HIVpol ORF was induced with 1 mM

isopropyl-P-D-thiogalactopyranosidefor 30minand labeled for 2 h withacombination of[35S]Met(1mCi) and[3H]Leuor[3H]Val(5mCi).Bacterialextracts

were prepared and fractionatedby RPHPLCasdescribed above. Selected isopropanol gradient fractions were immunoprecipitated with pooledserafrom AIDSpatients and resolvedbySDS-PAGE.Theputativeprotease(arrow)waselectroeluted, and its N-terminalsequence wasdetermined(seethelegendtoFig.2);theresiduesidentifiedareindicated by thecaretswithin the relevant region of the deducedsequence of theHIVpolORF belowtheautoradiogram.Thenumbers abovetheproteinsequencerefertothe residues(1to1012)of the HIVpol ORF.

were contiguous and if the protease were cleaved after

Arg-155 (denoted by! inFig. 2b).Afragment of this sizewas not observed among the tryptic digestion products.

How-ever, the experimentally determined positions of Leu and Val withina large peptide, L10 (probably overlapping

Arg-155), coincidedwiththeirpositionsinthepol ORFbetween Lys-138 and Arg-155. Among different HIV isolates and otherretroviral proteases, there isaconserved domain of 6 to9 residues surrounding an invariant Arg atposition 155. We believe, therefore, that this Arg might reside within a

hydrophobic pocket and be shielded from tryptic attack. Since both the viral and bacterial proteins immunoreacted with antiserum raisedagainsta 14-residue peptidebetween

residues154and 167of thepol ORF, the C terminus of the protease is most likely contiguous with the N terminus of RT.Onthisbasis the HIV proteasewasassumedtocontain 99 amino acids witha calculated molecular mass of10,774

andtoberelativelyrichinbasic residueswithacalculated

pl

of 9.83. It had a canonical -DTG- sequence (conserved

among all retroviral gag proteases and generic aspartyl proteinases) atposition 25. Theprotein washighly

homolo-gous to retroviral proteases over a 13-residue region

cen-teredaround the -DTG- sequence.Two other lesserregions ofhomology with other retroviral proteases were centered around amino acidpositions52and 87(correspondingtothe polORF residues 120 and 155, respectively).

Thefollowing three types ofmechanisms have been de-scribed for theexpressionofthe

pol

ORF of retroviruses: (i) suppression ofan amber codon between the gag and

pol

ORFs of murine leukemia virus and feline leukemia virus

(21); (ii)asingleribosomalframeshiftnearthe start of the

pol

ORF of Rous sarcoma virus and HIV (10, 12); and (iii) doubleframeshiftingatthejunctionsof the

gag-Xlpro

andX/ pro-polORFs of mouse mammary tumor virus (11). Since

the HIV proteaseiscontainedwhollywithin the

pol

ORF, it

islikely thatframeshifting occurs eithernearthe beginning ofthe

pol

ORForimmediately upstream of theprotease N

terminus. Frameshiftingatthe former sitehas recentlybeen

Al 1A

A2 ₆₉

30 14

kDa

- 200

- 92.5

- 69.0

-46.0

-30.0

- 21.5

- 14.3

on November 10, 2019 by guest

http://jvi.asm.org/

[image:5.612.63.569.66.307.2]

demonstrated in vitro (10). Whatever the mechanism of gag-pol precursor synthesis, the initial cleavage of the protease moiety from the precursor is apparently an auto-catalytic process, asevidenced by the presenceof identical Ntermini for both the viral andE. coli-expressed proteins. Weare grateful to Malcolm A.Martin forsupport and encourage-ment. The help of Charles E. Buckler in computer analysis is acknowledged. Malcolm A. Martin and Arnold Rabson are also thanked forcriticalreview of themanuscript.

LITERATURE CITED

1. Adachi, A., H. E.Gendelman, S. Koenig, T. Folks, R. Willey, A. Rabson, and M. A. Martin. 1986.Production of acquired immu-nodeficiency syndrome-associated retrovirus inhumanand non-humancellstransfected withaninfectious molecular clone. J. Virol. 59:284-291.

2. Alexander, F., J. Leis, D. A. Soltis, R. M. Crowl, W. Danho, M. S. Poonian, Y.-C. E.Pan, and A. M. Skalka. 1987. Proteo-lytic processing of aviansarcomaandleukosis virusespol-endo recombinant proteins revealsanotherpolgenedomain.J. Virol. 61:534-542.

3. Coligan,J. E., F. T. Gates III, E. S.Kimball,and W. L.Maloy. 1983. Radiochemical sequence analysis ofbiosynthetically la-beled proteins.MethodsEnzymol. 91:413-444.

4. Crawford, S.,andS. P.Goff. 1985. Adeletion mutationin the 5' partofthepol geneof Moloney murine leukemia virusblocks proteolytic processing of thegagand pol polyproteins.J.Virol. 53:899-907.

5. Debouck, C., J. G. Gorniak, J. E. Strickler, T. D. Meek, B. W. Metcalf, and M. Rosenberg. 1987. Human immunodeficiency virus protease expressed in Escherichia coli exhibits autopro-cessing and specific maturation of the gag precursor. Proc. Natl.Acad. Sci. USA84:8903-8906.

6. Dittmar, K. J., and K.Moelling. 1978.Biochemicalpropertiesof p15-associatedproteaseinanavian RNA tumorvirus. J. Virol. 28:106-118.

7. Eisenman, R.N., W. S. Mason, and M. Linial. 1980. Synthesis andprocessing of polymerase proteins of wild-type andmutant avian retroviruses. J. Virol.36:62-78.

8. Farmerie, W. G., D. D. Loeb, N. C. Casavant, C. A. Hutchison III, M. H. Edgell, and R. Swanstrom. 1987. Expression and processing ofthe AIDS virus reverse transcriptase in Esche-richiacoli. Science 236:305-308.

9. Gendelman, H. E., T. S. Theodore, R. Willey, J. McCoy, A. Adachi, R. J. Mervis, S. Venkatesan, and M. A. Martin. 1987. Molecular characterization of a polymerase mutant human immunodeficiencyvirus. Virology 160:323-329.

10. Jacks, T., M. D. Power, F. R. Masiarz, P. A. Luciw, P. J. Barr, and H. E. Varmus. 1987. Characterization ofribosomal

frame-shiftingin HIV-1gag-polexpression.Nature(London) 331:280-283.

11. Jacks, T., K. Townsley, H. E. Varmus, andJ. Majors. 1987. Two efficient ribosomalframeshiftingevents arerequiredfor synthe-sis ofmousemammary tumorvirusgag-related polyproteins. Proc. Natl.Acad. Sci. USA 84:4298-4302.

12. Jacks, T., and H. E. Varmus. 1985. Expression ofthe Rous sarcomaviruspolgenebyribosomalframeshifting. Science230: 1237-1242.

13. Lightfoote, M. M., J. E. Coligan, T. M. Folks, A. S. Fauci,M.A. Martin, and S. Venkatesan. 1986.Structural characterizationof reverse transcriptase and endonuclease polypeptides of the acquired immunodeficiency syndrome retrovirus. J. Virol. 60: 771-775.

14. Lillehoj, E. P., N. B. Myers, D. R. Lee, T. H.Hansen, andJ.E. Coligan. 1985.Structural definition ofafamily of Ld-like mole-cules distributedamongfourofsevenhaplotypescompared.J. Immunol. 135:1271-1275.

15. Rice, N. R., R. M.Stephens, A. Burny, and R. V.Gilden.1985. The gag and pol genes of bovine leukemia virus: nucleotide sequence andanalysis. Virology 142:357-377.

16. Robey, W. G., B. Safai, S. Oroszlan, L.0. Arthur, M. A. Gonda, R. C. Gallo, and P. J. Fischinger. 1985. Characterization of envelope andcore structural geneproducts of HTLV-III with serafromAIDSpatients. Science 228:593-595.

17. Steimer, K.S., K. W. Higgins, M. A. Powers, J. C.Stephans, A. Gyenes, C.George-Nascimento, P. A.Luciw, P. J. Barr,R. A. Hallewell, and R. Sanchez-Pescador. 1986. Recombinant poly-peptide fromtheendonuclease region of the acquiredimmune deficiency syndrome retrovirus polymerase (po) gene detects serumantibodiesinmostinfected individuals.J.Virol.58:9-16. 18. Veronese, F. D., T. D.Copeland, A. L. DeVico, R. Rahman, S. Oroszlan,R. C.Gallo, and M. G.Sarngadharan. 1986. Charac-terization ofhighly immunogenic p66/p5l as the reverse tran-scriptase of HTLV-III/LAV. Science 231:1289-1291.

19. Vogt, V. M., A. Wight, and R. Eisenman. 1979. Invitrocleavage of avianretrovirusgagproteins by viralproteasep15. Virology 98:154-167.

20. Witte,0.N., and D. Baltimore. 1978.Relationship of retrovirus polyprotein cleavagestovirionmaturation studied with temper-ature-sensitivemurineleukemia virusmutants.J.Virol. 26:750-761.

21. Yoshinaka, Y.,I. Katoh, T. D. Copeland, and S. Oroszlan. 1985. Murineleukemia virusproteaseis encoded by the gag-polgene and issynthesized throughsuppression ofanambertermination codon. Proc. Natl. Acad. Sci. USA 82:1618-1622.

22. Yoshinaka, Y.,I. Katoh, T. D. Copeland, G. W. Smythers, and S. Oroszlan. 1986.Bovine leukemiavirusprotease:purification, chemical analysis, and in vitro processing of gag precursor polyproteins.J. Virol. 57:826-832.

on November 10, 2019 by guest

http://jvi.asm.org/